Terrestrial Activity and Conservation of Adult California Red-Legged Frogs Rana Aurora Draytonii in Coastal Forests and Grasslands

Biological Conservation 110 (2003) 85–95 www.elsevier.com/locate/biocon

Terrestrial activity and conservation of adult California red-legged frogs Rana aurora draytonii in coastal forests and grasslands

John B. Bulgera, Norman J. Scott Jr.b,*, Richard B. Seymourc a580 Swanton Road, Davenport, CA 95017, USA bUS Geological Survey, Western Ecological Research Center, Piedras Blancas Field Station, Post Oﬃce Box 70, San Simeon, CA 93452, USA c818 Hollenbeck Ave., Sunnyvale, CA 94087, USA

Received 5 December 1999; received in revised form 15 January 2002; accepted 1 March 2002

Abstract The federally threatened California red-legged frog Rana aurora draytonii occupies both aquatic and terrestrial habitats in its adult life stage. The terrestrial activities of this species are not well known and require documentation to assist in the development of appropriate levels of protection under the US Endangered Species Act. We studied the terrestrial activities of radio-tagged red- legged frogs (n=8–26) inhabiting a coastal watershed in Santa Cruz County, California, during 1997–1998. In particular, we investigated (1) the use of terrestrial habitats by non-migrating adults in relation to season, breeding chronology, and precipitation, and (2) adult migration behavior, including seasonal timing, duration, distances traveled, and the use of corridors. Non-migrating red-legged frogs occupied terrestrial habitats briefly (median=4–6 days) following infrequent summer rains, but resided nearly continuously on land (median=20–30 days) from the onset of the winter wet-season until breeding activities commenced 1–2 months later. All of the non-migrating frogs remained within 130 m of their aquatic site of residence (median <25 m). Intervals spent on land were again brief during mid/late winter (median=1–4 days), despite frequent and copious rainfall. Adult migration to and from breeding sites occurred from late October through mid-May (wet season). We monitored 25 migration events between aquatic sites that were 200–2800 m apart. Short distance movements (<300 m) were completed in 1–3 days, longer movements required up to 2 months. Most migrating frogs moved overland in approximately straight lines to target sites without apparent regard to vegetation type or topography. Riparian corridors were neither essential nor preferred as migration routes. Frogs tra- veling overland occurred in upland habitats as far as 500 m from water. Approximately 11–22% of the adult population was estimated to migrate to and from breeding sites annually, whereas the bulk of the adult population was resident at these sites. Adequate protection of red-legged frog populations inhabiting relatively undeveloped landscapes is liable to be achieved by retaining an array of suitable habitat elements within at least 100 m of occupied aquatic sites, and by imposing seasonal limitations on detrimental human activities occurring within this zone. Specific protections for migrating frogs are probably unwarranted in forest and rangeland environments because dispersal habitat is ubiquitous and migrating frogs are widely distributed across the landscape in space and time. Published by Elsevier Science Ltd. Keywords: California red-legged frog; Amphibian migration; Buffer zones; Wildlife corridors; Amphibian conservation

1. Introduction habitat patches (Saunders and Hobbs, 1991, Hobbs, 1992, Beebee, 1996, Semlitsch, 1998). Many anurans Applied conservation strategies for threatened and would appear to be appropriate candidates for this type endangered species frequently rely on the use of buﬀer of approach. One such species, the California red-legged zones around critical habitat elements and on connect- frog Rana aurora draytonii, was recently listed as feder- ing corridors that allow for dispersal between suitable ally threatened (USFWS, 1996), and is now a focal species in habitat protection and mitigation programs associated with land-use planning and the exploitation * Corresponding author. Present address: Herpetology, Natural History Museum of Los Angeles County, 900 Exposition Boulevard, of natural resources. Los Angeles, CA 90007, USA. Tel.: +1-805-227-4246. California red-legged frogs occur from sea level to 1500 E-mail address: [email protected] (N.J. Scott Jr.). m elevation and occupy a variety of aquatic habitats in

0006-3207/02/$ - see front matter Published by Elsevier Science Ltd. PII: S0006-3207(02)00179-9 86 J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 landscapes that are used predominately for agriculture, Winter temperatures rarely drop below 3 C, and sum- ranching, timber harvesting, or recreation. They are mer temperatures seldom exceed 32 C. Morning and largely absent from urban and suburban residential set- evening fog is common during the summer months. tings. Important wetland sites for the species are rather Annual rainfall is usually in the range 1.0–1.5 m, with easily identified and some form of protection or man- 90% of it falling between November and May. In this agement of these wetlands theoretically preserves or study, El Nin˜ o weather patterns prevailed during the enhances their habitat value to frogs. By contrast, little 1997–1998 wet season, during which 1.8 m of rainfall is known about the terrestrial activities of red-legged was recorded in our study area, including nearly 1.0 m frogs in relation to any upland vegetation type or land during January and February. use category. Stebbins’ (1966) assessment that the species is ‘‘generally found in or near water but disperses after rains’’ adequately characterizes our current under- 3. Methods standing of seasonal habitat use. In consequence, there is insufficient information on which to base decisions 3.1. Study species regarding land-use planning, land management, and the regulation of activities in terrestrial environments in a California red-legged frogs are active the year-round manner that achieves appropriate levels of protection of in coastal areas. The species breeds in ponds, marshes, this species and its habitats under the US Endangered and lentic waters of streams between December and Species Act of 1973. April (Jennings and Hayes, 1995). When not breeding, The purpose of the present study was to gather data red-legged frogs occupy a broad spectrum of wetland on terrestrial activity by adult California red-legged habitats, including natural and artificial ponds and frogs living in a coastal landscape dominated by forests reservoirs, streams and other watercourses, freshwater and grasslands. Using radio transmitters, we investi- lagoons, springs, and seeps. Males mature in 2 years, gated (1) the use of upland habitats in relation to sea- females in 3 years (Jennings and Hayes, 1985). Adult son, breeding chronology, and precipitation, and (2) males (78–116 mm) are smaller than adult females (87– migration characteristics including timing, duration, 138 mm) (Hayes and Miyamoto, 1984). distance traveled, and the use of corridors. Because we were not testing any specific hypotheses, the study is of 3.2. Capture and radio transmitter attachment a descriptive nature and is intended to provide a quantitative basis for the development of management and We captured frogs at night and processed and regulatory strategies that take into account both the released them onsite within 10–20 min after capture. aquatic and terrestrial habitat requirements of this species. Apart from trying to maintain equal sample sizes of both sexes, we selected individuals for radio-tagging haphazardly. Sites where frogs were radio-tagged (see 2. Study area Fig. 3) were chosen for their proximity to coniferous forests and included both ponds (n=4) and streams The study area is located in the lower Scott Creek (n=3 sites). All study ponds were <0.1 ha in surface watershed on the western slope of the Santa Cruz area. Mountains, northern Santa Cruz County, California Radio transmitters were attached to frogs using a (see Fig. 3). Elevations range from sea level to 300 m. bead-chain belt that was secured around the waist Aquatic habitats include man-made stock ponds and (Rathbun and Murphey, 1996). We used Holohil model agricultural reservoirs, permanent and seasonal streams, BD-2G transmitters, designed with a 20 week battery and a coastal lagoon at the mouth of Scott Creek. Sev- life and weighing 1.65 g. The total weight of each unit, eral vegetation types are represented, but at larger scales including the waist band, was less than 2.0 g. Only adult the area is dominated by grassland and scrub on the frogs (males with thumb pads, females >88 mm) were coastal terrace to the west of Scott Creek, and by con- tagged. Adults weighed 48–214 g, with males smaller iferous forest from Scott Creek eastward. Coast red- than females. Radio-tagged frogs also were injected wood Sequoia sempervirens predominates in drainage with PIT tags (Camper and Dixon, 1988) for individual bottoms and on north-facing slopes, whereas various identification in the event of recapture after transmitter mixes of redwood, Douglas-fir Pseudotsuga menziesii, failure or loss. Monterey pine Pinusradiata , and hardwoods occur elsewhere. Permanent streams frequently support a nar- 3.3. Radio-tracking schedule row fringe of riparian woodland composed chiefly of red alder Alnusrubra . Data on radio-tagged frogs were gathered from 16 The area has a moderate Mediterranean climate, with May 1997 through 22 May 1998. We attempted to wet mild winters and relatively dry, temperate summers. maintain long-term records on 15–20 individual frogs, J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 87 but there was considerable attrition as individuals shed home site at the end of the upland interval. We refer to transmitters or transmitters failed, so new individuals these as non-migrating frogs. Some individuals addi- were added as necessary. At any given time 8–26 adult tionally made overland movements between two aquatic frogs carried radio transmitters. In all, 56 radio-tagged sites, typically before or after breeding. We refer to individuals contributed to the data reported on here. these as migrating individuals. This dichotomous char- We present time series data in this paper by individual acterization refers to the behavioral event itself and is tracking sessions. Each tracking session comprised an not a permanent label assigned to the individual. Data interval of 1–3 days during which all tagged frogs were on use of terrestrial habitats by migrating and non- precisely located once. The average number of days migrating frogs are treated separately in the analyses taken to complete a tracking session was 1.4. Generally, below (Sections 4.1 and 4.2). three sessions were completed per week (mean=2.8 ses- Frogs occupying terrestrial habitats were hidden from sions/week). In total we conducted 143 radio-tracking view in 96% of the tracking observations (n=961). Live sessions. Because we did not track frogs every day, plants (shrubs/herbs) provided concealing cover in 79% durations of individual bouts of terrestrial activity often of the cases, followed in descending importance by were imprecisely known (Æ1 or 2 days). Thus, median woody debris and rootballs (14%), small recesses in values reported here for various types of terrestrial vertical banks (4%), and forest floor litter (3%). The activity are often stated as a minimum–maximum range. most commonly used cover species included California For most analyses, small sample sizes preclude any blackberry Rubusursinus , poison oak Toxicodendron meaningful breakdown by sex. diversilobum, and coyote brush Baccharispilularis , all of which were ubiquitous throughout the study area. 3.4. Data records 4.1. Part A: non-migrating frogs All radio tracking was done during daylight hours, when adults are generally inactive (Hayes and Tennant, 4.1.1. Description of terrestrial habitat use 1985). Frogs occupying terrestrial habitats were located Use of terrestrial habitats by non-migrating frogs by homing in on and then circling the tagged individual showed a clear response to rainfall during the summer at a distance of a meter or less. Because frogs on land and early winter months (Fig. 1). Frogs were virtually were nearly always hidden beneath concealing cover (see always <5 m from their pond or stream of residence below), close-range tracking very seldom disturbed a during dry intervals of the summer, but moved outward focal animal. into upland habitats to distances of up to 130 m in Each time a frog was located its position was mapped response to summer rain (n=8–18 individuals). With (1:12,000 scale) with reference to a magnetic compass the onset of the more frequent rains of winter, the bearing and distance from a known landscape feature. median distance from water increased to a relatively Data were recorded on its behavior, proximity to water constant 15–25 m through mid-December (n=6–13 (pond, stream, or ravine), general vegetation type, and individuals). Some frogs remained at distances of up to specific cover type. Distances relating to proximity to 100 m from water until late January. Ninety percent of water were estimated visually or paced if they were <50 the non-migrating frogs were always within 60 m of m, and paced if they were in the range 50–200 m. These water (Fig. 1). were rounded to the nearest 10 m over the range 30–100 No similar response to rainfall was evident during m and to the nearest 25 m over the range 100–200 m. mid- to late winter. Red-legged frogs (n=8–24) made Longer distances were paced or measured from mapped little use of upland habitats during this season despite locations and rounded to the nearest 50 m. In measur- frequent and copious rainfall (Fig. 1). The reduced use ing distances traveled by migrating frogs, movement of uplands appeared to be primarily related to breeding was considered to have been in a straight line between chronology. All radio-tagged males (n=4) entered points of contact with the frog. breeding ponds and began calling from late November Daily rainfall and temperature records were obtained through mid-December. Six of seven radio-tagged from a privately operated recording station located near females entered breeding ponds from early January the center of the study area. through mid-February, and spawning in these ponds occurred from mid-January through late-March. The sharp drop in the median distance from water during 4. Results mid-December and the correspondingly sharp drop in the maximum and 90th percentile distances from water Adult red-legged frogs occupied terrestrial habitats in late in January (Fig. 1) reflect the timing of these events. two distinct ways. Most individuals made short-range Individuals of both sexes remained closely tied to aquatic forays into upland habitats for periods of days to weeks environments as the breeding season progressed. From in response to precipitation but returned to their aquatic February through May, 90% of the non-migrating frogs 88 J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95

Fig. 1. Time series plots showing use of terrestrial habitats by non-migrating red-legged frogs in relation to rainfall, Santa Cruz County, California. Data are plotted by individual radio-tracking sessions for (A) median, 90th percentile, and maximum distances from water, and (B) daily rainfall totals. Sample sizes of tagged non-migrating frogs ranged from 8 to 24. were always within 6 m of water. The sample of tagged the mid- to late winter season (8.2 C, range=2.2– frogs over this interval included up to 11 males and 13 13.9 C), when they did not. females. Variability in the ambient air temperature within the 4.1.2. Duration of boutson land study area was so slight during the winter months that it Table 1 shows the duration of continuous intervals was unlikely to have been a factor in frog use of uplands. individuals spent in upland habitats at distances of >10 The mean daily low temperature during the early winter m from water. Seasonal cutoﬀs were assigned based on season (9.0 C, range=2.8–16.1 C), when frogs made obvious changes in rainfall/frog response (Fig. 1). The extensive use of uplands, was nearly identical to that of distance criterion of 10 m was chosen to eliminate the J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 89

Table 1 150 m from the home pond before returning to it. Small Duration of continuous intervals that non-migrating red-legged frogs sample sizes and imprecise medians preclude a statistical spent in terrestrial habitats (>10 m from water) by season, June 1997– comparison between the two data sets. May 1998, Santa Cruz County, California

Season Month No. Median Maximum 4.2. Part B: migrating frogs initiated bouts duration duration a (days) (days) We monitored 25 migration events, including 10 from Summer June–October 31 4–6 22 sites of summer residence to breeding ponds (breeding Early Winterb November–December 18 20–30 >63 migration) and six elective movements from breeding Mid/Late Winter January–May 9 1–4 7–10 ponds to sites of summer residence (post-breeding a Medians are given as ranges because frogs were not tracked every migration). The remaining nine events were movements day. Thus, most bouts were recorded to Æ1 or 2 days. away from a pond that dried during June 1997 and b Of the 18 bouts, seven were truncated because of predation, trans- again during May 1998. mitter failure, lost contact with the frog, or having tagged the frog sub- sequent to its initiation of the interval on land. The broad median takes these into account. 4.2.1. Timing of breeding migration We recorded the starting date for 9 of the 10 breeding migration events (Fig. 2). Eight of nine frogs migrating influence of background levels of upland occurrence, as to breeding ponds initiated the movement between 31 frogs routinely occupied a fringe of land immediately October and 25 November. The ninth individual, a adjacent to water regardless of ambient conditions (see female, delayed its departure from its summer habitat Fig. 1). During the summer season, when frogs moved until 10 January. The first frog to migrate did so prior into upland habitats in response to widely separated to the onset of winter rains and traveled via a ravine single rain events, the median duration of intervals that held running water over most of its length. Initia- spent in these habitats was 4–6 days (maximum=22 tion of the remaining eight breeding migration events days). By contrast, with the more regularly spaced rains was in each case associated with a 1-day rainfall total in of early winter, the median duration of bouts in uplands excess of 25 mm. Although rainfall clearly facilitates was in the range 20–30 days (maximum >63 days). movements, it is possible that frogs migrating to breed- During mid/late winter, bouts on land were again brief ing sites are not hormonally prepared to move until (median=1–4 days), despite the frequent occurrence of sometime late in October. No migration was associated rain. with 8 mm of rain on two days early in October, nor with 28 mm late in August, despite the fact that frogs 4.1.3. Use of ravines were able to reside in terrestrial habitats for up to 22 Contiguous with the ponds at two of our radio tag- days in the wake of these rains (Table 1). ging sites were ravines that had at least some surface flow during most of the year. Roughly one-third of all 4.2.2. Timing of post-breeding migration non-migrating frogs at these sites spent at least some Post-breeding migration occurred between 29 January time in ravines during each season (summer=1 of 3 and 1 May, and in each of six cases was initiated in frogs; early winter=3 of 9; mid/late winter=7 of 23). association with a one-day rainfall total of from 10 to Data on ravine bout-duration and distances moved 40 mm (Fig. 2). For females, there was some overlap in away from the home pond (Table 2) were comparable to the timing of breeding and post-breeding migration. The the corresponding data for movements into upland first female to depart a pond after breeding (29 January) habitats (Fig. 1, Table 1). Non-migrating frogs inhab- did so two weeks prior to the arrival of the last female (13 ited ravines for as long as 38–42 days and moved up to February) at the same pond. Our interpretation of these

Table 2 Use of ravines by non-migrating red-legged frogs at two sites where ravines with surface ﬂow were connected to study ponds, June 1997–May 1998, Santa Cruz County, California

Season Date No. Median Maximum Median distance Maximum distance initiated bouts duration duration (m) from Ponda (m) from Pond (days) (days)

Summer June–October 1 14–15 n/a n/a 40 Early Winter November–December 4 21–25 38–42 55 65 Mid/Late Winter January–May 11 3-7 24–26 30 150

a Median of the maximum distance each individual moved from the pond. 90 J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95

Fig. 2. Timing and duration of individual red-legged frog migration events in relation to daily rainfall totals, 1997–1998, Santa Cruz County, California. Dashed lines indicate that the event was truncated due to predation or loss of the radio transmitter. Abbreviations are M=male, F=female. One frog (female 15919) was excluded due to uncertainty regarding the onset of the event. She arrived at the breeding pond on 3 January. Asterisks show the initiation timing of each migration event.

movements as ‘‘post-breeding’’ is supported by the facts (500 m), including both recently-tilled fields and matur- that (1) none of the three males that migrated from a ing crops. breeding site did so until after the last known clutch of The duration of individual movements between eggs was laid, and (2) all three females that migrated aquatic sites ranged from 1–3 days at a minimum to 57– from breeding sites had lost 15–28% body mass during 60 days at a maximum (Table 3). Although there was their tenure in the pond, suggesting that each had considerable variability in durations of longer move- spawned. ments (>500 m), distance moved and movement duration were highly correlated (rs=0.901, n=16, 4.2.3. Migration distance, duration, and use of uplands P<0.0001). Migrating frogs moved between sites that were sepa- Thirteen of 16 migrating frogs traveled primarily or rated by map distances of 200–2800 m. Each segment of exclusively through terrestrial habitats between sites, a migratory route took the frog closer to its eventual covering total overland distances of 400–3200 m for target. Thus, movements were highly oriented toward completed movements. The longest single overland seg- target sites and usually tended to describe an approxi- ment traversed without contacting a pond or stream was mately straight line between the source and target sites 1200 m. Migrating frogs temporarily occupied upland (Figs. 3 and 4). The minimum distance actually traveled habitats as far as 500 m from any aquatic site (Table 3). was in most cases close to the shortest straight-line dis- Most migrating individuals moved to the nearest tance (Table 3). The longest route traveled was 3600 m pond to breed, and to the nearest pond or stream after by an individual moving between two sites 2800 m apart. breeding (Table 3). The three exceptions to this pattern The tendency of most individuals to move in all resided in the same pond (not used for breeding) approximately straight lines to target sites suggests de during the summer and then migrated to breed in a facto that there was neither avoidance of nor preference pond 2800 m distant. In this case, the nearest known for any particular landscape feature or vegetation type. breeding pond was 600 m from the site of summer resi- Frogs traveled readily through the range of upland dence (Fig. 3). habitats represented in the study area. Maximum dis- Streams in the study area are not known to be used tances moved by individual frogs through various for breeding, presumably because scouring peak flows macro-habitat types were: coniferous forest (900 m), coincide with the red-legged frog breeding season. There grass/scrub rangeland (1700 m), and agricultural land are virtually no off-channel backwaters on these J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 91

streams. Six of seven frogs tagged in streams moved to ponds to breed (Table 3). The seventh was a female that had sustained a major abdominal wound, presumably during a predation attempt. She did not breed, and remained throughout the winter at the stream.

Fig. 3. Map of the study area, northern Santa Cruz County, Cali- fornia, and routes (arrows) taken by red-legged frogs migrating from sites of summer residence to breeding sites (breeding migration). A star indicates that the frog was preyed upon or shed its transmitter prior to reaching the target site. Circles represent ponds. Ponds used as radio- tagging sites appear as hollow circles (n=4). Permanent and seasonal streams are shown as solid and dashed lines. Locations of stream seg- ments used as radio-tagging sites are indicated by two parallel lines running through a solid circle (n=3). The shaded area shows the Fig. 4. Routes (arrows) taken by red-legged frogs migrating from extent of the coniferous forest. Unshaded areas are either grassland breeding sites to sites of summer residence (post-breeding migration), and scrub or agricultural ﬁelds. northern Santa Cruz County, California.

Table 3 Summary data related to red-legged frog breeding and post-breeding migration events, 1997–1998, Santa Cruz County, California (Values in par- entheses indicate truncated events.)

Frog Map distance Minimum Movement Total Longest Maximum End Nearest between distance duration overland overland distance point pond or sites moved (days) distance segment from water habitatsc stream? (m) (m) (m) (m)a (m)b

Breeding migration 15919 200 200 1–3 unk. unk. unk. P to P Yes E755B (300) (350) (2–4) (350) (350) (250) S to P Yes 46118 (500) (550) (24–26) 550 550 150 S to P Yes C2302 (550) (550) (13) 0 0 <10 S to P Yes 81F54 650 800 17–19 800 800 300 S to P Yes 02555 650 800 39–43 800 800 250 S to P Yes F5C45 700 750 20–23 400 400 150 S to P Yes 1105C (2600) (2600) 57–60 (2600) 750 350 P to P No 22E79 2800 3200 53–56 3200 1200 500 P to P No 3432B 2800 3600 35–38 1150 600 300 P to P No Post-breeding migration 15919 200 200 1–3 unk. unk. unk. P to P Yes 51C77 200 200 2–3 unk. unk. unk. P to P Yes 81F54 300 350 2–4 350 350 150 P to S Yes 22E79 450 500 14–16 500 500 225 P to P Yes B0B45 500 500 2–4 500 500 250 P to P Yes F5C45 700 700 16–20 0 0 <10 P to S Yes

a Longest distance traveled between two points of contact with water. Value is 0 if frog moved via a ravine. b Maximum distance from a pond, stream, or ravine with surface ﬂow. c P=Pond; S=Stream. 92 J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95

4.2.4. Topographic component of migration days (early June–late July) before entering the nearest The study area is comprised of low mountainous ter- pond, and one moved overland 300 m to the nearest rain that is bisected at regular intervals by steep-sided stream over the course of 9–11 days. drainages. The average percent slope for paired aquatic sites between which frogs migrated ranged from 12 to 4.2.8. Estimate of adult migration rate 33%. Ignoring the averages, canyon side-slopes fre- Based on prior observations, we did not expect red- quently exceeded 50% slope over distances of 100 m or legged frogs to breed at stream sites or at one of our more. One such slope, which was scaled by two tagged study ponds. In consequence, frogs tagged at these sites frogs, included a 77% elevation gain over 180 m. Except during the summer of 1997 had been selected, in part, for vertical rock faces, we did not observe any topo- with an intentional bias toward a high likelihood of graphic constraints to frog migration. migration, in order to gather data pertinent to that life- The three frogs that made the longest movements history characteristic. Of the original batch of 10 frogs each traveled via a different route to reach the same that migrated to breeding sites, we were able to main- destination (Fig. 3). The longest of these followed drai- tain contact with four through May 1998. We also nage contours to the maximum extent possible, travel- attached a radio transmitter to a male captured on land ling 3600 m in the horizontal plane and 190m in while in the process of moving 500 m to a breeding elevation (minimum possible). The shortest of the trio pond. We followed this individual through May 1998 as made a nearly straight-line movement of 2800 m in the well. All five of these individuals also migrated post- horizontal plane, but traveled approximately 610 m breeding, three to their previous summer site and one to upward and downward in elevation by crossing topo- a new site; the origin of the fifth frog was unknown. All graphic contours over five drainage divides. five had migrated by 1 May. By contrast, three other individuals that were followed from the summer of 1997 4.2.5. Ratesof travel during migration through May 1998 did not migrate either pre- or post- Migrating frogs tended to move in spurts of a day or breeding. Thus, breeding and post-breeding migrations two and then remain quiescent for intervals of several may involve predominantly the same individuals. days. Because we did not locate individual frogs daily, To obtain an unbiased estimate of the proportion of we do not have tightly bounded data on numbers of adults that migrate to and from breeding ponds, we days spent moving versus resting. Known distances attached radios to 20 haphazardly selected individuals moved (>30 m) in 24 hours (n=8) ranged from 150– at two breeding sites (Site 1=5 males and 4 females; Site 500 m. 2=5 males and 6 females). At each site the first 4–6 frogs of each sex encountered during capture sessions 4.2.6. Use of corridors for migration were fitted with radios. The experimental group was In the study area, watercourses (streams and ravines) tracked from early February through early May, which and their associated vegetation are the only obvious encompassed the range of post-breeding migration dates features of the landscape that might function as migra- in the control group. tion corridors for migrating red-legged frogs. The Of the 20 selected frogs, two migrated to new sites, migration routes of 14 individuals were documented in one was preyed upon early in April, and two shed their sufficient detail to evaluate the role of watercourse cor- radios during April and could possibly have migrated ridors in facilitating migration. Of these 14 individuals, subsequently. Excluding the individual lost to preda- six inhabited sites with no obvious connecting corridor tion, the proportion of migrating individuals repre- to the target site, and thus could only (and did) travel sented in this sample (n=19) falls in the range 10.5– via overland routes. By contrast, eight frogs (at three 21.1%. (Data for each site are 1–3 of 9, and 1 of 10). sets of paired sites) had the immediate option to migrate These data suggest that there is a relatively small seg- to the target site via a connecting watercourse route. Of ment of the adult population that is liable to migrate in the eight, three principally followed watercourses and any given year, and that most adults are resident the five traveled via overland routes. Thus, obvious migra- year around at favorable breeding sites. Morphological tion corridors in the form of watercourses or riparian characteristics of frogs that migrated were indis- vegetation strips were neither essential to facilitating tinguishable from those that did not. The mean snout- red-legged frog migration nor were they used in pref- urostyle length of males that migrated was 88.2 mm erence to overland routes. (n=5, range=81–96 mm) compared with a mean of 87.9 mm (n=8, range=79–95 mm) for those that did not 4.2.7. Movementsaway from drying ponds (two-tailed t=0.102, df=11, P=0.921). For females, the We monitored nine individuals that abandoned a corresponding data were mean length of 106.0 mm (n=7, pond that dried during early June 1997 and early May range=88–117 mm) for those that migrated, versus 1998. Seven of the nine moved to the nearest permanent 104.9 mm (n=10, range=88–123 mm) for those that pond 200 m away, one lingered in a wet ravine for 51–52 did not (two-tailed t=0.191, df=15, P=0.851). J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 93

5. Discussion be moving overland between aquatic sites at any time during the winter months (tentatively mid-October This study provides the first quantitative description of through mid-May). However, given (1) the compara- terrestrial habitat use by California red-legged frogs and, tively low numbers of individuals involved in migration as such, has important conservation implications pertain- (<25% of the adult population), (2) that individuals ing to the development of appropriate land management move to a particular site over a broad spatial scale, and strategies and regulatory guidelines for this species. In (3) that migration is spread out over time and does not particular, we are able to make biologically-based recom- occur as a synchronous en masse event, the density of mendations for the application and design of terrestrial red-legged frogs migrating through uplands is usually buffer zones at occupied aquatic sites, and to evaluate the so low that protective considerations may often be role corridors play in adult red-legged frog migration. unwarranted. This conclusion, however, may pertain only to rela- 5.1. Conservation of non-migrating frogs: buffer zones tively undisturbed environments, such as forests and rangeland, where artificial barriers to migration and Data on migration rates from this study indicate that persistent human-related sources of mortality are lar- more than 75% of the adult population is resident at gely absent. High-quality dispersal habitat is nearly permanent aquatic sites over the course of a year. ubiquitous in these types of landscapes. In degraded Moreover, 90% of the radio-tagged frogs that were not environments, this general conclusion may not apply. migrating between aquatic sites remained within 60 m of As landscapes become increasingly developed with water at all times and the farthest any non-migrating buildings, abnormally high predator densities, roads, frog moved from water was 130 m. Although a larger and related infrastructure, connectivity between aquatic sample size might enhance the precision of these dis- sites decreases and migration between aquatic habitat tance thresholds, our results are liable to be quite robust patches may become more perilous (Simberloff et al., given that there was little variability in observed beha- 1992; Rosenberg et al., 1997). Low recruitment of dis- vior among individuals. Non-migrating frogs uniformly persing individuals is likely to play an insidious and showed no proclivity to wander far from the home site, primary role in the extirpation of frog populations from even during the early winter season (n=238 tracking suitable aquatic sites in developing landscapes (sensu records; 17 frogs) when they spent continuous intervals Sjo¨ gren, 1991; Sinsch, 1992b; Sjo¨ gren Gulve, 1994; Sta- of up to 2 months on land (Table 1). This is particularly cey et al., 1997; Vos and Chardon, 1998). noteworthy in view of our finding that red-legged frogs are capable of moving 150–500 m in a single night. 5.3. The role of corridorsin red-legged frog migration These data indicate that the judicious application of terrestrial buffer zones adjacent to small ponds and Where an obvious, direct corridor exists between two streams may often be an effective means of protecting occupied aquatic sites, it undoubtedly will receive reg- and maintaining populations of California red-legged ular use by migrating frogs. However, there is no evi- frogs in coastal forests and grasslands (see also Rudolph dence from this study that natural corridors are either and Dickson, 1990; Burke and Gibbons 1995; Dodd, essential to migrating frogs or that they will be used 1996; Dodd and Cade, 1998; Semlitsch, 1998). Conserva- preferentially over alternative upland routes. This find- tion and resource management planning for activities ing in no way diminishes the value of ravines and small that alter the local environment should strive to retain a watercourses as frog habitat per se, and they constitute well-distributed array of natural habitat elements that important landscape features for red-legged frogs the provide protective cover for red-legged frogs to a distance year around (Table 2), regardless of their relative role in of at least 100 m from occupied aquatic sites. Dense pat- facilitating migration. ches of shrubs and herbaceous vegetation are particularly Attempts to mitigate adverse impacts to red-legged important habitat elements in this regard. Within desig- frog habitat through the designation or creation of nated buffer zones, incidental mortality of red-legged movement corridors in areas scheduled for development frogs resulting from human activities occurring adjacent are problematical. The tendency of frogs to move in to occupied wetlands can likely be minimized or avoided more or less straight lines to target sites indicates that it through seasonal restrictions. Our data indicate that the would be difficult to attempt to channel movements potential for detrimental impacts to red-legged frog through provisional corridors (see also Rathbun et al., populations is highest during the early winter months. 1997). Moreover, local topography, vegetation, and drainage contours cannot necessarily be used to predict 5.2. Conservation measures relating to migrating frogs likely migration routes to or from occupied sites. In our study, three frogs that moved between two aquatic Due to overlap in the timing of breeding and post- sites that were 2800 m apart did so over a swath breeding migration events (Fig. 2), individual frogs may through the landscape that was 600–700 m wide 94 J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95

(Fig. 3); one individual followed drainages to the max- 5.5. Limitationsof the data imum extent possible, whereas two moved overland across the topography. Additionally, although many Our estimate of adult red-legged frog migration rates individuals move between adjacent aquatic sites, not all is derived from frogs that inhabit permanent, small do. Thus, proximity to another suitable site cannot ponds (<0.1 ha) that support breeding and year-round always be used to predict directional dispersal prob- use and where nearby alternative aquatic sites are abilities. Our prediction for the three individuals men- available. Whereas these conditions are not uncommon tioned above would have been that they would migrate elsewhere within the species’ range, the migration rate at 600 m northward to the nearest breeding pond, rather any particular site is liable to be a function of many than 2800 m southward to an alternative site. variables, including rainfall patterns, fluctuations in Recent reviews of the literature on wildlife corridors water levels, degree of isolation from other sites (Sjo¨ g- have concluded that there is yet little evidence in sup- ren, 1991; Sjo¨ gren Gulve, 1994), regional metapopula- port of their effectiveness in achieving desired goals, tion density, and carrying capacity of the site. Similarly, despite their widespread use in conservation strategies we note that patterns of terrestrial habitat use by non- (Simberloff et al., 1992; Rosenberg et al., 1997; see Beier migrating frogs that inhabit large, complex wetland sites and Noss 1998 for dissenting view). While we suggest may be substantially different from those reported here. caution in the application of corridors as a mitigation The extent to which our results can be generalized tactic for red-legged frogs, our results may not be awaits data from other study sites. wholly applicable to developed landscapes. It is possible This study describes behavior and habitat use of adult that, as habitat connectivity decreases, red-legged frogs frogs only. Subadult red-legged frogs occupy a range of will become more reliant upon topographic or vegeta- aquatic habitats and also are found dispersing during tive corridors or that they may exhibit other compensa- rains, but are not obliged to reside at breeding sites during tory behavioral responses that favor successful the winter. From routine capture efforts, we know that migration (e.g. Rosenberg et al., 1998). subadults are under-represented at breeding ponds during the winter season (personal observation). Presumably 5.4. Anuran migration: comparative data they are primarily terrestrial at this time, but we lack data on use of terrestrial habitats for this age class. This Aspects of adult migration have been investigated in a includes an appreciable proportion of the population, small number of anuran species and in general the data including males <2yearsandfemales<3 years of age compare favorably with those presented here on Cali- (Jennings and Hayes, 1985). Post-metamorphic anurans fornia red-legged frogs. In our study, adult red-legged may often disperse radially from their natal sites (Dole, frogs traveled map distances of up to 2.8 km in a single 1971; Schroeder, 1976; Sinsch, 1997), and are known to season. Among the more vagile anuran species, adult move up to 5 km over the 2–3 years between metamor- Bufo calamita and B. bufo are known to make seasonal phosis and first-breeding (Schroeder, 1968, 1976; Dole, migrations of up to 3 km (Sinsch, 1989), Hoplobatrachus 1971; Berven and Grudzien, 1990; Sinsch, 1997). There is occipitalis of up to 6 km (Spieler and Linsenmair, 1998), convincing empirical evidence that post-metamorphic and Rana lessonae of typically 1 km or less (Sjo¨ gren dispersal contributes significantly more to regional meta- Gulve, 1994). In a review of data for 16 species of North population persistence than does adult dispersal (Sinsch, American anurans, Dodd (1996) reported movement 1992b; Sinsch and Seidel, 1995; Sinsch, 1997; also see distances of 1–2 km in Bufo americanus, B. cognatus, Breden 1987). Thus, allowance for juvenile dispersal may Rana capito, and R. pipiens. The straight-line move- in some cases be a critical consideration in long-term ments and spatial orientation toward target sites that we conservation planning. Application of our results to real recorded for red-legged frogs have also been docu- conservation situations should beware this limitation. mented in other species of frogs and toads (van Gelder et al., 1986; Sinsch, 1989, 1990, 1992a; Kusano et al., 1995; Sjo¨ gren Gulve, 1998; Matthews and Pope, 1999). Acknowledgements Rates of travel, too, are similar among species studied. Our radio-tagged red-legged frogs moved 150–500 m Funding for this study was provided by the US Fish overnight, compared with 89–440 m for B. bufo (Git- and Wildlife Service’s Sacramento and Ventura offices, tens et al., 1980; van Gelder et al., 1986), 120–1400 m and we wish to thank Steve Morey and Cathy McCalvin for H. occipitalis (Spieler and Linsenmair, 1998), and for their efforts in this regard. We are additionally up to 160 m for Rana pipiens (Dole, 1965). During grateful to Bud McCrary and Wally Mark for permis- migration, brief bouts of movement separated by sion to work on lands owned, respectively, by Big Creek longer intervals of inactivity typify both red-legged Lumber Company and California Polytechnic State frogs (this study) and common toads (van Gelder et al., University, San Luis Obispo, and to Lud and Barbara 1986). McCrary for providing rainfall and temperature records J.B. Bulger et al. / Biological Conservation 110 (2003) 85–95 95 from the study area. Laura Perry, of the Land Trust of Canyon National Park, California. Journal of Herpetology 33, Santa Cruz County, was instrumental in administering 615–624. many aspects of the project and provided some of the Rathbun, G.B., Murphey, T.G., 1996. Evaluation of a radio-belt for ranid frogs. Herpetological Review 27, 187–189. field equipment. Thanks are due also to Galen Rathbun Rathbun, G.B., Scott, N.J. Jr, Murphey, T.G., 1997. Rana aurora and Tom Murphey, USGS-BRD, Piedras Blancas, for draytonii (California red- legged frog): behavior. Herpetological early help and advice with the field work, to Lynn Review 28, 85–86. Rathbone for the maps, and to Susan Wright for expe- Rosenberg, D.K., Noon, B.R., Meslow, E.C., 1997. Biological corri- diting the transfer of materials from the office to the dors: form, function, and efficacy. BioScience 47, 677–687. Rosenberg, D.K., Noon, B.R., Megahan, J.W., Meslow, E.C., 1998. field. The manuscript benefited from editorial comments Compensatory behavior of Ensatina eschscholtzii in biological corri- by Stephen Corn, Kenneth Dodd, Gary Fellers, Steve dors: a field experiment. Canadian Journal of Zoology 76, 117–133. Morey, Galen Rathbun, Ulrich Sinsch, and an anon- Rudolph, D.C., Dickson, J.G., 1990. Streamside zone width and amphi- ymous reviewer. Our sincere thanks to each of them. bian and reptile abundance. Southwestern Naturalist 35, 472–476. Saunders, D.A., Hobbs, R.J. (Eds.), 1991. The Role of Corridors. Surrey Beatty & Sons, Chipping Norton, NSW, Australia. References Schroeder, E.E., 1968. Movements of subadult greenfrogs, Rana clamitans. Journal of Herpetology 1, 119. Beebee, T.J.C., 1996. Ecology and Conservation of Amphibians. Schroeder, E.E., 1976. Dispersal and movement of newly transformed Chapman & Hall, London. green frogs, Rana clamitans. American Midland Naturalist 95, 471– Beier, P., Noss, R.F., 1998. Do habitat corridors provide connectivity? 474. Conservation Biology 12, 1241–1252. Semlitsch, R.D., 1998. Biological definition of terrestrial buffer zones for Berven, K.A., Grudzien, T.A., 1990. Dispersal of the wood frog (Rana pond-breeding salamanders. Conservation Biology 12, 1113–1119. sylvatica): implications for genetic population structure. Evolution Simberloff, D.S., Farr, J.A., Cox, J., Mehlman, D.W., 1992. Move- 44, 2047–2056. ment corridors: conservation bargains or poor investments. Con- Breden, F., 1987. The effect of post-metamorphic dispersal on the servation Biology 6, 493–504. population genetic structure of Fowler’s toad Bufo woodhousei Sinsch, U., 1989. Migratory behaviour of the common toad Bufo bufo fowleri. Copeia 1987, 386–395. and the natterjack toad Bufo calamita. In: Langton, T.E.S. (Ed.), Burke, V.J., Gibbons, J.W., 1995. Terrestrial buffer zones and wetland Amphibians and Roads. ACO Polymer Products Ltd, England, pp. conservation: a case study of freshwater turtles in a Carolina Bay. 113–124. Conservation Biology 9, 1365–1369. Sinsch, U., 1990. Migration and orientation in anuran amphibians. Camper, J.D., Dixon, J.R., 1988. Evaluation of a microchip marking Ethology, Ecology & Evolution 2, 65–79. system for amphibians and reptiles. Texas Parks and Recreation Sinsch, U., 1992a. Sex-biased site fidelity and orientation behaviour in Department, Research Publication 7100–159, 1–22. reproductive natterjack toads (Bufo calamita). Ethology, Ecology & Dodd, C.K. Jr, 1996. Use of terrestrial habitats by amphibians in the Evolution 4, 15–32. sandhill uplands of north central Florida. Alytes 14, 42–52. Sinsch, U., 1992b. Structure and dynamic of a natterjack toad meta- Dodd, C.K. Jr, Cade, B.S., 1998. Movement patterns and the con- population (Bufo calamita). Oecologia 90, 489–499. servation of amphibians breeding in small, temporary wetlands. Sinsch, U., 1997. Postmetamorphic dispersal and recruitment of first Conservation Biology 12, 331–339. breeders in a Bufo calamita metapopulation. Oecologia 112, 42–47. Dole, J.W., 1965. Summer movements of adult leopard frogs, Rana Sinsch, U., Seidel, D., 1995. Dynamics of local and temporal breeding pipiens, in northern Michigan. Ecology 46, 236–255. assemblages in a Bufo calamita metapopulation. Australian Journal Dole, J.W., 1971. Dispersal of recently metamorphosed leopard frogs, of Ecology 20, 351–361. Rana pipiens. Copeia 221–228. Sjo¨ gren, P., 1991. Extinction and isolation gradients in metapopula- Gittens, S.P., Parker, A.G., Slater, F.M., 1980. Population character- tions: the case of the pool frog (Rana lessonae). Biological Journal istics of the common toad (Bufo bufo) visiting a breeding site in mid- of the Linnean Society 42, 135–147. Wales. Journal of Animal Ecology 49, 161–173. Sjo¨ gren Gulve, P., 1994. Distribution and extinction patterns within a Hayes, M.P., Miyamoto, M.M., 1984. Biochemical, behavioral and northern metapopulation of the pool frog, Rana lessonae. Ecology body size differences between Rana aurora aurora and R.a. draytonii. 75, 1357–1367. Copeia 1018–1022. Sjo¨ gren Gulve, P., 1998. Spatial movement patterns in frogs: target- Hayes, M.P., Tennant, M.R., 1985. Diet and feeding behavior of the oriented dispersal in the pool frog, Rana lessonae. Ecoscience 5, 31–38. California red-legged frog Rana aurora draytonii (Ranidae). South- Spieler, M., Linsenmair, K.E., 1998. Migration patterns and diurnal western Naturalist 30, 601–605. use of shelter in a ranid frog of a West African savannah: a tele- Hobbs, R.J., 1992. The role of corridors in conservation: solution or metric study. Amphibia-Reptilia 19, 43–64. bandwagon? Trends in Ecology and Evolution 7, 389–392. Stacey, P.B., Johnson, V.A., Taper, M.L., 1997. Migration within Jennings, M.R., Hayes, M.P., 1985. Pre-1900 overharvest of Cali- metapopulations: the impact upon local population dynamics. In: fornia red-legged frogs (Rana aurora draytonii): the inducement for Hanski, I., Gilpin, M.E. (Eds.), Metapopulation Biology: Ecology, bullfrog (Rana catesbeiana) introduction. Herpetologica 41, 94–103. Genetics, and Evolution. Academic Press, San Diego, pp. 267–291. Jennings, M.R., Hayes, M.P., 1995. Amphibian and Reptile Species of Stebbins, R.C., 1966. A Field Guide to Western Reptiles and Amphi- Special Concern in California. Report prepared for the California bians. Houghton Mifflin, Boston. Department of Fish and Game, Inland Fisheries Division, Rancho U.S. Fish & Wildlife Service, 1996. Determination of threatened status for Cordova, CA. the California red-legged frog. Federal Register 61 (101), 25813–25833. Kusano, T., Kazuko, M., Kaneko, S., 1995. Post-breeding dispersal of van Gelder, J.J., Aarts, H.M., Staal, H.W.M., 1986. Routes and speed the Japanese toad, Bufo japonicusformosus . Journal of Herpetology of migrating toads (Bufo bufo L. ): a telemetric study. Herpetologi- 29, 633–638. cal Journal 1, 111–114. Matthews, K.R., Pope, K.L., 1999. A telemetric study of the Vos, C.C., Chardon, J.P., 1998. Effects of habitat fragmentation and movement patterns and habitat use of Rana muscosa, the road density on the distribution pattern of the moor frog Rana mountain yellow-legged frog, in a high-elevation basin in Kings arvalis. Journal of Applied Ecology 35, 44–56.